mycroft

Use mycroft to ingest EPUBs and ebooks, build a local vector index, and ask questions about a book.

Setup (once) - export OPENAI_API_KEY="..." - mycroft config onboard - mycroft config resolve

Common commands - List books: mycroft book list - Ingest EPUB: mycroft book ingest /path/to/book.epub - Ingest with summaries: mycroft book ingest /path/to/book.epub --summary - Ingest with batch embeddings (50% cheaper): mycroft book ingest /path/to/book.epub --batch - Ingest with batch summaries + embeddings: mycroft book ingest /path/to/book.epub --batch --summary - Resume batch ingestion: mycroft book ingest resume <id> - Check ingestion status: mycroft book ingest status <id> - Show metadata: mycroft book show <id> - Ask a question: mycroft book ask <id> "What is the main conflict?" - Search passages: mycroft book search <id> "mad hatter" --top-k 5 - Delete book: mycroft book delete <id> --force - Start chat: mycroft chat start <id> - Ask in session: mycroft chat ask <session> "What does this foreshadow?" - Continue chat: mycroft chat repl <session>

Notes - Use mycroft config path to find the config file location. - book ask and book search require embeddings and an OPENAI_API_KEY. - Chat commands require embeddings and an OPENAI_API_KEY. - Prefer book search and synthesize answers yourself before using book ask. - Summaries increase ingestion time and cost significantly; enable --summary only when needed. - Use --batch to run embeddings and summaries via the OpenAI Batch API at 50% cost; results may take up to 24 hours. When combined with --summary, summaries are batched first, then embeddings on resume. - After --batch ingestion, use mycroft book ingest status <id> to check progress and mycroft book ingest resume <id> to complete indexing. - If a non-batch ingest is interrupted, use mycroft book ingest resume <id> to continue from the last saved chunk. - If a batch fails, resume automatically re-submits it. - For scripted runs, avoid interactive flags like --manual or omit confirmations with --force.